Humble

Did you solve that ?

a small bug? ```python text = '全国一共有112所211大学..' from paddlespeech.t2s.frontend.zh_frontend import Frontend as zhFrontend fe = zhFrontend() print(sum(fe.get_phonemes(raw_text), [])) # Outputs: [全国一共有一百一十二所二幺幺大学..] not in g2pW dict,use g2pM ['j', 'ie2', 'k', 'e4',...

And what if I just train on ONE scene ? Can I extract all object radiance fields (maybe in different category) in the trained multi-view scene ?

@Jeffwan same problem as you, do you have any idea to solve this?

> @QinlongHuang Make sure your batch setting is correct. You can check more details here #188. Thx for your reply ! But it not works for me... It does work...

@Jeffwan Thx for your patient answer, but I found it is a problem about nccl backend of 4090 card. @lksysML With setting `NCCL_P2P_DISABLE=1`, I finally solve this problem. See #250...

直接改asrmodel中相应的维度就可以了

您说的效果不好是指模型run不起来还是训完结果不太好呢？我也是在hf上随便找了个model, jonatasgrosman/wav2vec2-large-xlsr-53-chinese-zh-cn

NOT totally solved. I find the problem was due to NCCL backend trying to use peer to peer (P2P) transport. So setting envionment variable `NCCL_P2P_DISABLE=1` or `NCCL_P2P_LEVEL=2` might fix the...

> @QinlongHuang I can confirm NCCL_P2P_DISABLE=1 fixed issues with my 4x 4090 on epyc! Also nvidia driver with cuda 12.1 has not yet fixed this NCCL issue with 4090. Glad...